Table 2.

Diagnostic performance of the hybrid model for the prediction of IDH status

Dataset	Accuracy	AUROC (95% CI)			AUPRC (95% CI)
Per slice
Internal test set	90.50%	0.96	0.95	0.97	0.88	0.82	0.93
SNUH set	84.90%	0.93	0.9	0.95	0.81	0.74	0.87
TCIA set	77.10%	0.84	0.81	0.86	0.79	0.74	0.83
Per patienta
Internal test set	93.80%	0.96	0.93	0.99	0.88	0.72	0.98
SNUH set	87.90%	0.94	0.89	0.97	0.82	0.65	0.94
TCIA set	78.80%	0.86	0.8	0.91	0.81	0.71	0.88

AUROC, area under the receiver operating characteristics curve; AUPRC, area under the precision-recall curve.

^aSince each patient yielded 5 tumor slices, the diagnostic accuracy per patient was calculated from the mean value of the 5 predicted probabilities per patient.